AI Models Collapse When Trained on Recursively Generated Data¶
Shumailov, I., Shumaylov, Z., Zhao, Y., Papernot, N., Anderson, R., & Gal, Y. (2024). AI Models Collapse When Trained on Recursively Generated Data. Nature, 631, 755-759.
Cited by¶
1 citation across 1 artifact.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Deception Blowback
- In AI training, synthetic data injected to fill a gap re-enters a later model's training set as ground truth, compounding errors at scale.
This sourceDemonstrates model collapse when synthetic/model-generated outputs re-enter later training corpora as if authentic, causing irreversible degradation — the self-poisoning blowback of unsegregated generated data re-ingested by a later training run.
- In AI training, synthetic data injected to fill a gap re-enters a later model's training set as ground truth, compounding errors at scale.
Verification¶
This reference passed the adversarial substantiation pipeline: it was checked to exist and to support the claim it is attached to. See how references were verified.
Registry ID ref:1bd0f3e40333 · see in the full table